DIMICOLI AND BIETH
Location of the Calcium Ion Binding Site in Porcine Pancreatic Elastase Using a Lanthanide Ion Probe Jean Luc Dimicoli*J and Joseph Bieth
ABSTRACT: Calcium and gadolinium ions are able to bind to elastase at identical sites and with similar affinities. The dissociation constants of the elastase calcium and gadolinium complexes (4.5 X and 2 X M, respectively, at pH 5 and 35 "C) are not significantly affected by the presence of a specific inhibitor of this enzyme, trifluoroacetyltrialanine. Conversely, the binding of calcium to elastase does not modify the enzyme's affinity for the inhibitor, nor its reactivity toward its specific substrate, succinyltrialanine-p-nitroanilide. The ~ MT 2 E I M of the fluorine nuclei of the relaxation times T ~ E and trifluoroacetyltrialanine in the ternary elastase-inhibitorgadolinium ion complex have been obtained by factorization of the observed relaxation times in terms of exchange and true relaxation contributions. The distance from the gadolinium ion to the trifluoroacetyl group of the inhibitor has then been
c
alcium ion binding on serine proteases involves specific sites of the proteins. One calcium ion is bound to bovine trypsin in the crystal at a site which includes the Glu-70' and the Glu-80 side chains (Bode and Schwager, 1975). This site is different, however, from that assigned for the binding of a calcium ion to bovine trypsin (Abbott et al., 1975; Darnall et al., 1975) and chymotrypsin A (Birnbaum et al., 1977) in solution. These authors conclude from fluorescence energy transfer and nuclear magnetic resonance (NMR)2 experiments using lanthanide probes that the calcium ion binding site in solution is on the side chains of Asp-I94 and Ser-190. They reported also that pancreatic elastase has a very low affinity for a calcium ion, probably due to the lack of the Ser-190 hydroxyl group in this enzyme (Darnall et al., 1975). The Glu-70 and Glu-80 residues, however, are still present in this protease and they bind to a uranyl ion in the crystal (Shotton and Watson, 1970). As will be shown in this paper, the above conclusion of Darnall et al. on elastase is not correct: calcium, as well as gadolinium ions, form stable complexes with elastase. Moreover, the lanthanide ion may be used as a paramagnetic probe for exploring the vicinity of the ion binding site at the surface From the Centre de Recherche Delalande, 92 500, Rueil-Malmaison, France, the Fondation Curie-Institut du Radium, 91405 Orsay, France, and the Laboratoire de Chimie Biologie, U.E.R. de Sciences Pharmaceutiques, 67083 Strasbourg, France. Received March 25, 1977. This work is supported by the Dtltgation G&n&alei la Recherche Scientifique et Technique (Contract No. 76-7-1857). Address correspondence to this author at the Foundation Curie-Institut du Radium, Throughout this paper, we shall use both the chymotryptic numbering system for the amino acid sequence of serine proteases (Stroud et al., 1971) and the subsite nomenclature of Schechter and Berger ( I 967) for the active site of proteases. Abbreviations used are: NMR, nuclear magnetic resonance; Tris, tris(hydroxymethy1)aminomethane; EDTA, ethylenediaminetetraacetic acid.
* '
5532
BIOCHEMISTRY, VOL.
16,
NO.
25, 1977
calculated to be 20 A by using the Solomon-Bloembergen equations. This rather large distance explains in part the independence of the two fixation sites of the ion and the inhibitor. The previous demonstration that the trifluoroacetyl group binds in the vicinity of one of the S sites, together with crystallographic coordinates of pancreatic elastase, permits us to conclude that the fixation site of the calcium ion to elastase in solution is most probably the same as that of a uranyl ion to this enzyme in the crystal, i.e. the carboxylic side chains of Glu-70 and Glu-80 (based on the chymotrypsin sequence numbering system). In addition, the rate constant koff of dissociation of the elastase-peptide complex, measured under a large concentration range, is almost constant, thus suggesting that the trifluoroacetyltrialanineis bound in a unique mode to the enzyme.
of the protein. We used for this latter purpose the N M R of the fluorine label of a trifluoroacetylated trialanine peptide, which has been previously shown (Dimicoli et al., 1976) to form a stable complex with elastase. Materials and Methods
Elastase and Trifuoroacetyltrialanine. Porcine pancreatic elastase and trifluoroacetyltrialanine were prepared and assayed as previously reported (Bieth et al., 1974; Dimicoli et al., 1976). The various kinetics and inhibition parameters and their 95% confidence intervals were obtained using a program of nonlinear regression. Enzyme-Ion Binding Equilibria. Enzyme-ion binding was measured by equilibrium dialysis under the following condiM acetate buffer at pH tions: 2 X M protein, 5 X 5 , or 0.1 M Tris buffer at pH 8, and various concentrations of NaCI. Such conditions prevent the Donnan effect from effecting the relative ion concentrations in both compartments of the cell. On the other hand, it can be easily shown that the effect of complex formation of lanthanides with acetate ion (Sonesson, 1958), in large and constant concentration in the medium, is limited to the competition of the acetate ion with the enzyme, leading to a lower apparent affinity of the enzyme for the lanthanide ion. In addition, approximately IO-* M 45CaC12(IRE, Fleurus, Belgium) was present in each dialysis cell. After equilibration at 35 or 5 "C during a 15- 18-h period, the emissions from the two compartments of each cell were monitored using a Intertechnique SL32 scintillation counter. The difference between the numbers of counts emitted by equal volumes of the compartments containing and not containing the enzyme, CI and Cz, respectively, proved the existence of binding between the 45Ca2+ion to the enzyme. Moreover, as the concentration of the 45Ca2+ion is much lower than that of the free enzyme, this latter concentration, E , can be obtained from the relation: Ko(Ci - C2) E= (la)
c2
C A L C I U M BINDING S I T E IN E L A S T A S E
where KO is the overall dissociation constant of the elastase 45Ca2+ion complex, and is obtained in the absence of any other added ion. For this latter condition, the relation l a is still valid, with the total enzyme concentration, eo, appearing in the place of the concentration E . On the other hand, the total quantity of the nonradioactive ions (Ca2+ or (id3+) was determined a t the beginning of the experiments by weighing the CaC12 and by complexometric titrations by EDTA using xylenol orange (Lyle and Rahman, 1963) as an indicator in 0.2 M acetate buffer (pH 5 . 6 ) , for gadolinium nitrate5HzO. The free enzyme concentration, determined by counting the 45Ca2+ion containing solutions, and the total nonradioactive ion concentration permit us to evaluate the free, and enzyme-bound, nonradioactive ion concentrations. A Scatchard plot (Scatchard, 1949) can then be constructed that gives an initial estimation of the number of sites of fixation of each ion on the enzyme and the corresponding affinity constant. For a unique site of fixation which was always the case in the present experiment, the affinity constant could be refined and its 95% confidence interval obtained by a nonlinear regression analysis of the experimental data using the expression giving the concentration of the enzyme-bound ion, E M : EM =
Eo
+ Mo + K - [(Eo + Mo + K ) 2 - 4E0Mo]l/~ 2
(1b) where Mo and K are the total ion concentration and the dissociation constant of the enzyme-ion complex, respectively. N M R . The N M R spectra were recorded at 100 (IH) and 94 MHz (19F) using a Varian XLlOO spectrometer operating in the Fourier transform at a sample temperature of 34 "C. The chemical shifts were referenced to the resonance of trifluoroacetic acid in a coaxial capillary. The elastase solutions in D20-5 X M acetate buffer were prepared without NaC1, ensuring complete solubility of M. Spin-lattice relaxation times, T I , the enzyme up to were measured by the induction recovery method (Pratt and Sykes, 1972), using 180 OC-7-90 OC-t sequences. Spin-spin relaxation times, T2, were measured from the half-height line , an appropriate width of the resonance line (T2 = 1/ T A U ) using spectral resolution. No oxygen effects requiring systematic degassing of the solutions were observed for these measurements. In enzyme-inhibitor systems in which the chemical exchange is fast with respect to the difference in chemical shift for the observed nucleus of the inhibitor free in solution (I) and complexed to the protein (EI), i.e. when a single line is observed, the relaxation rates are given by the following relations (Smallcombe et al., 1972):
In these relations, Ti (i , 1,2) is the observed relaxation time, Til and T i ~ are l those for the free and complexed inhibitor, respectively, and PI, PEI,WI, and WEI are the relative proportions and frequencies of the nuclear precession in the two forms, respectively. It should be noted that the last term in eq 3a, which represents the contribution of the chemical exchange to the line width, is maximum for P I = 2/3. The dissociation rate constant of the complex, koff, may be derived from these relations once P I and P E I have been determined from the
measurements of the chemical shift of the observed resonance: 6 = pi61 + P E I ~ E I (4a)
To obtain P I and PEIfrom relation 4a, a trial starting value of 6 ~ is1 chosen that allows a first estimation of the number of fixation sites for the binding of the inhibitor to the enzyme, and of the corresponding dissociation constants, using a Scatchard representation. In the case of a unique site of fixation, these first estimations of the chemical shift, 6 ~ 1 and , of the dissociation constant, Kd, can be refined and their 95% intervals obtained by a nonlinear regression analysis of the experimental data using the expression giving the observed chemical shift:
-
X ( ~ E I 61)
(sa)
where IO is the total concentration of the inhibitor. The ratio k,ff/Kd gives, finally, the second-order rate constant k,, of formation of the enzyme-inhibitor complex. The analytical expression for the relaxation times T I and T2 of a nucleus in an inhibitor able to exchange chemically between three sites, the free inhibitor (I), the enzyme-inhibitor complex (EI), and the ternary complex with a metal ion (EIM), can be obtained through calculations similar to those used by Swift and Connick (1962) and Reuben and Fiat (1969). Under the following experimentally observed conditions-no change of chemical shifts upon addition of metal ( 6 ~ 1 = ~ E I M ) no , change in the dissociation and association rate constants, kerf and kon,for the enzyme-inhibitor complex upon addition of metal ion, fast chemical exchange rates relative to the relaxation times-relations 2b, 3b, and 4b hold: 1 =- PI
TI
1 - PI T2
PEI +-+-
I PEI I P E l M
(2b)
TII(M) T I E I T I E I M PElM
TZI(M) TZEI TZEIM PI'(PEI + P E I M ) ( ~ E I- 6 d 2 (3b) koff 6 = pi61+ (PEI+ P E I M ) ~ (4b) in which the index EIM refers to the ternary enzyme-inhibitor-metal complex, and T i 1 ( ~is) the relaxation time of the peptide in the presence of enzyme and ion. Equation 3b may be rewritten in a condensed form:
+
where 1/T2,exchis the contribution of chemical exchange to 1 /T2 corresponding to the last term of eq 3b, which is the same as in the diamagnetic case (eq 3a) and 1/ T i includes only the contributions of the enzyme and paramagnetic ion to the relaxation. The extrapolated values of TI and 1/T2' for P E ~ M = 1 allow us to estimate T I E ~and M T ~ Eand ~M to derive the values of 7C and r , by using the following simplified SolomonBloembergen equations ignoring the contributions due to the modulation of the scalar interaction between the electron spins on the gadolinium ion and the nuclear spins:
x
[
BIOCHEMISTRY, VOL.
1
+37c + 1 +77c W,27,2
W&c2
16, NO. 25, 1971
]
(5b)
5533
DIMlCOLl AND BIETH
K ,of the Complexes of Ca2+ with Porcine Pancreatic Elastase and Other Serine Proteases.
TABLE I: The Dissociation Constants,
Enzyme
pH
Temp ("C)
Porcine elastase
5 5 5 8 8 6 8 6.3
35 35 35 35 4 25 25
Bovine trypsin Porcine trypsin (I
This work.
Ionic strength
Effector
K (M)
Ref
0.2 0.05 0.05 0.2 0.2 0.3 0.3 0.2
0 0 Trifluoroacetyltrialanine Trifluoroacetyltrialanine 0 p-Toluamidine-chloride p-Toluamidine-chloride 0
1.4 f 0.3 X 4.5 f 0.5 x lo-' 4.3 f 0.5 x 10-5 2 i 0.7 X 7.3 f 0.6 x 10-5 3.8 x 10-3 3.0 x 1 0 - 4 1.6 x 10-4
a
5
Abbott et al. (1975).
TABLE 11: The Dissociation Constants,
Temp ("C)
Ionic strength
Effector
Porcine elastase Porcine elastase Porcine elastase Bovine trypsin Bovine chymotrypsin A Porcine trypsin
5 5 5 6 6 6.3
35 35 35 25 25 5
0.15 0.05 0.05 0.3 0.3 0.2
0 0 Trifluoroacetyltrialanine p-Toluamidine-chloride p-Toluamidine-chloride 0
Epstein et al. (1974).
0 0.5 1 Scatchard plot for the interaction of elastase with gadolinium M acetate buffer (pH 5) with ( 0 )and without (0)IO-' M NaCl at 35 "C, as obtained by competition with 45Ca2Cion.
[
37c
1
+
W[27,2
+
1
+137c
ws27,2
]
(5c)
where S is the electron spin quantum number for gadolinium (7/2), y1 is the fluorine magnetogyric ratio (2.518 X lo4 rad/(s G)), g is the electronic "g" factor (2), MB is the Bohr magneton (0.9273 1 X erg/G), w l is the Larmor angular precession frequency for nuclear spin of fluorine (5.91 X 1O'O rad/s), and w s is the Larmor angular precession frequency for electron spins (4.15 X I O 1 ' rad/s). Results (1) Binding of Ca2+ and Gd3+Ions to Elastase. The dissociation constants of the 45Ca2+elastase complex, measured by equilibrium dialysis under various conditions of pH, temperature, and ionic strength, and in the presence or absence of the trifluoroacetylated inhibitor, are reported in Table I. They
5534
C
BIOCHEMISTRY, VOL.
Kd
(M)
8.5 f 2 x 10-5 3.3 i 1 x 10-5 2 f 0.6 x IO-' 3 x 10-4 5 x 10-4 2.9 x Io-'
Ref a a U
b d C
Birnbaum et al. (1977).
FIGURE 1: ion in 5 X
x 47c+
a b b
K ,of the Complexes of Gd3+ with Porcine Pancreatic Elastase and Other Serine Proteases.
pH
Abbott et al. (1975).
U U
Epstein et al. (1974).
Enzyme
This work.
U
16, NO. 25, 1977
are comparable to those measured for other serine proteases, suggesting the presence of a specific site for this ion. Competition experiments with nonradioactive calcium ions confirm the uniqueness of this site. In addition, the presence of the inhibitor has no effect on calcium binding. The binding of trivalent gadolinium ion to elastase is also competitive with that of 45Ca2+ion (Figure 1). The affinity of elastase for this metal ion has been confirmed by direct titration of Gd3+ ion (Table 11). It is not affected by the presence of the elastase inhibitor and its value which is a minimal one due to the competition effect of acetate ion present in large excess (see above) is comparable or even higher than that for other serine proteases. (2) Effect of Ca2+and Gd3+Ions on the Properties of Elastase. Table 111 shows that Ca2+ and Gd3+ ions affect neither the kinetic parameters of the elastase-catalyzed hydrolysis of succinyltrialanine-p-nitroanilide,nor the dissociation constant of the elastase-trifluoroacetyltrialaninecomplex. These results are in agreement with the data of Tables I and I1 and demonstrate that the metal ion binding site and the substrate (inhibitor) binding site are quite different and independent. The rate of autolysis of the enzyme (ca. 20% after 15 h a t 35 OC and pH 5 ) is slowed down by the trifluoroacetyltrialanine inhibitor, but it is not further affected by the metal ions, thus confirming that the fixation of these ions and of the inhibitor is not mutually exclusive. (3) N M R Investigation of the Elastase-Tri'uoroacetyltrialanine Interactions. The 19Fresonance of this specific inhibitor of elastase has already been shown (Dimicoli et al., 1976) to be strongly dependent upon the presence of the enzyme. A quantitative analysis of its behavior in the ternary complex obtained with gadolinium ions first requires the estimation of the contribution of chemical exchange in the binary enzyme-inhibitor complex. The titration of the enzyme by the inhibitor results in a high field of the fluorine resonance with broadening, then narrowing, of the corresponding line. The maximum line width is observed for a p i value around */3 (Figure 2a), indicating the importance of chemical exchange in this process. The observed chemical-shift variations allow
CALCIUM
BINDING
SITE IN
ELASTASE
of the Calcium and Gadolinium Ions upon the Hydrolysis by Porcine Pancreatic Elastase (kcat, K,) of Succinyltrialanine-p-Nitroanilide,and Its Inhibition ( K Iby ) Trifluoroacetyltrialanine a t 35 O C and Various Conditions of pH and Ionic
TABLE 111: The Effect
Effector
Reaction
Ionic pH strength Temp ("C)
Succinyltrialanine-pnitroanilide
Hydrolysis
5
Trifluoroacetyltrialanine
Inhibition
5 5 5
Succinyltrialanine-pnitroanilide
Hydrolysis
0.05
35
0.05
8
0.05 0.05 0.2
35 35 35 25
8
0.2
25
K m (M)
Ion
Gd3+ (8 X
kcat
3 f 0.3
M) 2.3 f 0.3 X lo-)
3 f 0.3
0 Gd3+ (8 X
KI ( M )
(s-')
2.4 f 0.3 X IO-)
0
5.2 f 0.5 X 4 f 0.5 x 10-5
M)
0
1.2 f 0.3 X IO-)
17 f 2
Ca2+(10-3 M)
1 f 0.3 X IO-)
18 f 3
FIGURE 3: Scatchard plot for the interaction of elastase with trifluoroaM acetate buffer, pD 5, 35 OC, as measured cetyltrialanine in 5 X from the variations of chemical shift of the fluorine resonance of this peptide in various conditions of concentrations of enzyme and inhibitor.
0
0.0 5
01
p;(l-p,)
FIGURE 2: (a) Relaxation rate (T2-I - T I ' - ' )due to the chemical exchange for the fluorine resonance of trifluoroacetyltrialanineas a function ofpl, the fraction of inhibitor in the free state in the presence of 1.3 X M elastase in 5 X M acetate buffer, pD 5,35 O C . (b) Plot of (T2-I - T I - ' )(cf. a) vs.p12( I - P I )at pD 5 , in 5 X IO-* M acetate buffer, 35 O C . The 0 and points correspond to values ofpl respectively lower and higher than %.
+
us to initially obtain a Scatchard plot (Figure 3) showing that there is a single peptide inhibitor bound to the enzyme in this concentration range. The final value of the dissociation constant as obtained by nonlinear regression analysis, Kd = 1.3 f 0.4 X IOw4 M, is in good agreement with that measured by means of enzyme inhibition (Kd = 5 X M ) using much lower enzyme concentrations (1 0-8 M). The same regression analysis gives a fluorine chemical-shift variation in the complex of 149 f 6 H z at low-field values. Then p1 and p ~ can 1 be more precisely calculated for each measurement. It can then be seen that there is a linear variation o f ( T 2 - I - T1-I) as a function of p12(1 - PI) (Figure 2b), which would be expected from eq 2a and 3a when T l ~and l T ~ Eare I not significantly different. A value for koff of 590 s-I and of 4.4 X 1O6 M-I s-I for k,, has been derived from these relations. The times T ~ Eand I T ~ Eare I equal to 0.24 s. As already noted by Baldo et al. ( 1 9 7 9 , the value of k,, measured by N M R is far from that expected for a process controlled by diffusion and does not eliminate the possibility of stepwise binding, although the process is correctly
described in the whole concentration range using a single value for the rate constants. (4) NMR in the Ternary Elastase-Gadolinium(III)Peptide Complex. When a solution of the trifluoroacetylated peptide is titrated by gadolinium ion in the absence of the enzyme, the fluorine relaxation times TI and T 2 are shortened progressively. The ratio of their variations is constant and approximately equal to 7/6. Such a value corresponds to a correlation time T~ for the system dominated by the tumbling rate (Dwek, 1973) ( T
